Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: FLAC 1.3.0 has been released (Read 180344 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

FLAC 1.3.0 has been released

Reply #175
Didn't realize the AMD test set had valid tests for Intel too. On i7-4771:
flac -5 -s -c image.wav >nul:
oldsse 425.86x realtime
newsse 454.76x realtime

flac -8 -s -c image.wav >nul:
oldsse 176.06x realtime
newsse 177.82x realtime

FLAC 1.3.0 has been released

Reply #176
Too late for the game?
Where can I download flacABCD?

FLAC 1.3.0 has been released

Reply #177
They have been removed, as the tests were concluded and the relevant patch merged
Music: sounds arranged such that they construct feelings.

FLAC 1.3.0 has been released

Reply #178
Is flac_newsse now the fastest version for Intel i5-4670 cpu?
How about x32 vs x64?

FLAC 1.3.0 has been released

Reply #179
I still have the files from the tests, but since lvqcl removed them I have to assume he might not want them available here.

For me on Intel (i5-750) the fastest was flac64_d.
And 64bit was faster than 32bit.

FLAC 1.3.0 has been released

Reply #180
I have to assume he might not want them available here.

Those binaries should only ever be used for testing purposes. If you want a faster binary, *please* be patient and wait for the upcoming release, of which the source code will have been thoroughly tested.
Music: sounds arranged such that they construct feelings.

FLAC 1.3.0 has been released

Reply #181
Yes, I totally understand.
Just wish that I won't miss the party again.

FLAC 1.3.0 has been released

Reply #182
Yes, I totally understand.
Just wish that I won't miss the party again.

You don't have to be sad. Maybe you did miss that one http://www.hydrogenaud.io/forums/index.php?showtopic=106545
Many things you can benchmark till no sun shines. The forthcoming flac will be a real improvement.
Bench different apodizers, search methods, speed against compression, bring in flake and flacCL....
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

 

FLAC 1.3.0 has been released

Reply #183
32- and 64-bit exe files built from the current code (15 Oct 2014):

FLAC 1.3.0 has been released

Reply #184
Just wondering, is there a date/amount of new extras the code has to accumulate to warrant a new version?


FLAC 1.3.0 has been released

Reply #186
32- and 64-bit exe files built from the current code (15 Oct 2014):

Pink Floyd - Discovery Box Set (14 albums), converting with foobar2000 1.3.4 on i7-2700k

FLAC 1.3.0 (Case): Total encoding time: 3:36.563, 201.33x realtime

FLAC (flac-e0fbe71.exe, lvqcl) x64: Total encoding time: 1:24.969, 513.15x realtime

Wow, thanks very much for the patches and for integrating all the improvements, now let's wait for an official 1.3.1 release.

FLAC 1.3.0 has been released

Reply #187
Encoded all my Benassi CDs (3.11gb), on a laptop with SSD and i7-4700MQ, using Foobar.

1.3.0 john33 64bit: x255
git lvqcl 64bit: x630

Stunned.

FLAC 1.3.0 has been released

Reply #188
damn you all and your fancy cpu instructions.

on a single core athlon64 i get a 4% increase  (assuming the result is repeatable - i only tested one album)

Code: [Select]
flac 1.3.0 foobar2000 encoder pack level 8: Total encoding time: 2:09.781, 23.91x realtime
flac lvqcl level 8: Total encoding time: 2:04.625, 24.90x realtime

FLAC 1.3.0 has been released

Reply #189
damn you all and your fancy cpu instructions.
Indeed

3:23:55 of .tak files over gigabit LAN -> flac using foobar2000 with 4 threads:

Code: [Select]
flac-1.3.0: ~147,4x - 1:23 
flac-git:   ~254,9x - 0:48

CPU: Phenom II X4 970

edit: same set of files, but this time wav (ramdisk) -> flac (ramdisk)
Code: [Select]
flac-1.3.0: ~167,6x - 1:13 
flac-git:   ~339,9x - 0:36

FLAC 1.3.0 has been released

Reply #190
Here's my poor music playing little fanless SSD-equipped Atom at full steam:

flac.exe -f -8 The_Gathering-Echoes_Keep_Growing_-Edit-.wav

7.4x realtime (50.3 sec) with flac 1.2.1
8.4x realtime (44.6 sec) with http://www.saunalahti.fi/~cse/temp/flac-1.3.0-win32.zip
and a mighty
14.6x realtime (25.6 sec) with flac-e0fbe71.exe.7z

First and last repeated for sanity checking. Fairly stable.


flac.exe -t The_Gathering-Echoes_Keep_Growing_-Edit-.flac

Quite unstable results likely due to background processes, tested a handful of time each and reporting best results: respectively,
5.9 sec, 5.0 and 3.7 sec.


Stop whining over yer > 20x realtime figures! 
Kindly,
the fourth Yorkshireman

FLAC 1.3.0 has been released

Reply #191
However it uses much more RAM than before.

For instance, when encoding a 453 MB file using FLAC -6 with Flac-e0fbe71 by lvqcl then Timer32 reports:

Virtual  Memory  =    15 MB
Physical Memory =    12 MB

Whilst Flac 1.3.0 versions by Case, by LRN and by Rarewares are all reporting:

Virtual  Memory  =      4 MB
Physical Memory =      2 MB

FLAC 1.3.0 has been released

Reply #192
However it uses much more RAM than before.

Yes, that is correct. That is only the case with Windows though. It is an added buffer to reduce disk fragmentation.

See these thread for more information
http://lists.xiph.org/pipermail/flac-dev/2...ber/005081.html
http://lists.xiph.org/pipermail/flac-dev/2...ber/005094.html
Music: sounds arranged such that they construct feelings.

FLAC 1.3.0 has been released

Reply #193
Does the git version already include re-tuned  presets/new apodization functions? I'm seeing a minimal increase in filesize at -8

FLAC 1.3.0 has been released

Reply #194
Does the git version already include re-tuned  presets/new apodization functions?

No, it does not. I sent the patch for inclusion only yesterday.

edit: to be more specific, it does include the apodization functions, but they aren't used by default yet.
Music: sounds arranged such that they construct feelings.

FLAC 1.3.0 has been released

Reply #195
I see. I wonder why my 2GB set of wav files encoded with the git version are 3862 byte bigger than with 1.3.0 though. Not like it's a huge discrepancy


FLAC 1.3.0 has been released

Reply #197
I see. I wonder why my 2GB set of wav files encoded with the git version are 3862 byte bigger than with 1.3.0 though.

Probably the metadata written is slightly different or http://xiph.org/flac/faq.html#tools__different_sizes

I tried the old John33 64bit and lvqcl's 64bit compiles with my old frontend that writes own metadata and 30 cds came out exactly in the same size.
Again i am really impressed what you did to the code lvqcl alone from looking over the mailing list. A place in the flac Hall Of Fame is well deserved.
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

FLAC 1.3.0 has been released

Reply #198
Just thinking aloud:

For those of us who are a bit reluctant to using experimental builds for encodings meant to last forever: I assume there would be security added by decoding for verification with a different build? (Protecting against the case where new code makes "same" errors on encoding and decoding.) Is that necessary? Feasible?

FLAC 1.3.0 has been released

Reply #199
Just thinking aloud:

For those of us who are a bit reluctant to using experimental builds for encodings meant to last forever: I assume there would be security added by decoding for verification with a different build? (Protecting against the case where new code makes "same" errors on encoding and decoding.) Is that necessary? Feasible?

FLAC stores the MD5 hash of the original stream in its STREAMINFO block, so a broken encoder will manifest as a file that passes all the frame checksum verifications but has an MD5 mismatch at the end.  It should be easy enough to check the unlikely event of the decoder miscalculating the MD5 hash by piping its raw PCM output to an external md5sum program and comparing it to the contents of STREAMINFO.